An Improved Algorithm Based on Exemplar Based Image Inpainting

Abstract

Image restoration technology is an important image processing technology. In image restoration, it not only can reconstruct the loss part of impaired image caused by external factors, but also can remove unwanted parts from the specified image. Criminisi et al proposed an algorithm for image restoration processing, namely Exemplar Based Image Inpainting (EBII). Using the provided template window, this algorithm can find the best matching image data in the original image for filling after calculating the filling priority of the contour region of the image to be repaired. This approach can process various kinds of images effectively. However, in the treatment of some details in an image, the EBII algorithm needs to be improved. For instance, we discovered that the calculations of the priority and image matching would be influenced by the size of the template window, which was fixed in the EBII algorithm. Therefore, in this paper we present an improved EBII algorithm: a collective of template windows with various sizes will be used in image inpainting; in the repair process, image restoration is performed by judging the optimal template size in the collection of template windows. Our results show that the ameliorated EBII algorithm can significantly improve image quality owing to the delicate treatment of the details.